METHODS

The German CF quality assurance (CFQA) project is a patient registry which contains relevant clinical and laboratory data, respiratory function test results, and categorised CF treatment and complications.¹⁵ Data are entered into the database once a year for each patient. In 1997, clinical data from 3448 patients aged 0–58 years were available from 97 different CF outpatient clinics in Germany. For the present study the cohort of 3298 patients aged above 2 years of age was analysed.

Cross sectional analysis

The 1997 data from all patients were divided into four different age groups: 2–5.9 years, 6–11.9 years, 12–17.9 years, and 18.0 years of age or older, since severity of disease increases with age. According to a publication by Lai and coworkers,¹⁶ patients were defined as malnourished if data indicated wasting and/or stunting—that is, if (1) weight was below 90% of the predicted normal value for sex and height in children,¹⁷ if body mass index (BMI) was <19 kg/m² in adults or if (2) weight was <80% of the median normal value for sex and age, or if (3) height was <90% of the median normal value for sex and age.¹⁷ Pulmonary function was considered abnormal if (1) forced vital capacity (FVC) was <80% predicted and (2) if forced expiratory volume in 1 second (FEV₁) was below 75% of the predicted normal value.¹⁸ Mid expiratory flow at 25% of vital capacity (MEF₂₅) was rated abnormal if values were less than 60% predicted. The lower limit of normal for oxygen tension (Po₂) in oxygenised capillary blood was 10.7 kPa (80 mm Hg). Serum immunoglobulin G (IgG) values were considered increased if Z scores were ≥2.¹⁹ Reference values were used according to recommendations made by the Scientific Advisory Committee of the CFQA project who selected appropriate normal values for the population under consideration. The tabulated reference data for weight were related to a third degree polynomial equation of height. Comparisons between age groups were made for means of nutritional and lung function parameters, and the relative proportion of patients with abnormal results was determined for each age group. In addition, patients were grouped according to the presence or absence of risk factors such as Pseudomonas aeruginosa colonisation or malnutrition.

Multifactorial variance analysis was performed to investigate the association between lung function and malnutrition and the confounders P aeruginosa infection and sex. The Student's t test for independent samples was performed to compare means between groups with normal or abnormal results. χ² and Fisher's exact tests were used to evaluate proportions of patients.

Longitudinal analysis

Patients who had complete data sets for 1995 and 1996 were eligible for the first part of the longitudinal analysis. 536 children aged 6–11.9 years and 477 adolescents aged 12–17.9 years were analysed in detail. For each patient the differences in weight for height and FEV₁ between 1996 and 1995 were calculated.

Three longitudinal groups were defined according to the resulting differences: “better” if the results in 1996 were at least 5% higher than those in 1995: “no change” if the differences were between –5% and +5%; and “worse” if the results were at least 5% lower than in 1995. For each of these groups mean changes in FEV₁ and weight for height between the two years were determined. When means from the three longitudinal groups were compared, one way analysis of variance with the Student-Newman-Keuls test was performed. All calculations were made using SPSS (SPSS Inc, Chicago, USA).

In a second longitudinal analysis the equality of FEV₁ slopes and intercepts in different risk groups was tested for 1995, 1996, and 1997. Mixed model analysis was performed using SAS Proc Mixed version 6.12 for Windows.²⁰ A multiple comparison adjustment for the p values and confidence limits for the differences of least squares means was performed with a Tukey test. An approximate F test was used to assess the effect of deleting a variable from the model. A level of p=0.05 was used to indicate statistical significance.

RESULTS

Cross sectional data

Clinical and lung function data were available for 3298 patients above 2.0 years of age (table 1). As expected, mean weight for height, vital capacity, FEV₁, and Po₂ decreased with age, and the proportion of patients with abnormal results was larger in older patients. Mean weight for height was below 90% predicted in 19.0% of children aged 2–6 years and in 31.5% of adolescents. In children below 12 years of age, Po₂ had the highest frequency of abnormal results. More than 50% of adolescents had abnormal MEF₂₅, IgG, and Po₂, while FEV₁ was decreased in only 38.5%. A multifactorial analysis of variance revealed that FEV₁ was related to both malnutrition (p<0.001) and P aeruginosa colonisation (p<0.001) but was independent of sex (p>0.05) for all age groups.

Pulmonary function was decreased in 56.0% and 85.6% of P aeruginosa positive adolescents and adults, respectively, whereas only 37.4% and 66.3% of non-colonised patients had abnormal lung function (p<0.001, fig 1). An association between P aeruginosa colonisation and malnutrition was also observed. However, most patients had a weight for height above 90% predicted or a body mass index above 19 kg/m², irrespective of bacterial status.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Frequency of (A) malnutrition and (B) impaired lung function in patients with and without Pseudomonas aeruginosa (PA) colonisation.

Patients with malnutrition had significantly lower vital capacity and FEV₁ and higher serum IgG. As shown in table 2, this was independent of P aeruginosa infection. For example, in malnourished children aged 6–11.9 years who were P aeruginosa positive, mean vital capacity and FEV₁ were 11.3% and 14.5% predicted lower, respectively, than in normally nourished children (table 2). A similar difference between children of normal and decreased weight was observed in the absence of P aeruginosa colonisation—for example, FEV₁ was 11.8% predicted lower. In P aeruginosa positive and negative adolescents, malnutrition was associated with even larger differences in lung function, IgG, and Po₂. The differences between normally nourished and malnourished adults with CF were comparable to those in adolescents, and the presence or absence of P aeruginosa had no additional effect.

Figure 2 shows cross sectional mean values of FEV₁ by age and bacterial status. FEV₁ values of at least 90% predicted were observed in children and adolescents with normal weight and no P aeruginosa infection. P aeruginosa positive patients with normal weight had mean FEV₁ values which were 10–20% predicted lower, but they remained at above 80% predicted except at age 13. In contrast, malnutrition was associated with a severe decline in FEV₁ of about 20% predicted between 12 and 18 years of age, irrespective of bacterial status. Malnourished patients colonised with P aeruginosa had the worst results at any age. For example, at ages 8 and 17 mean FEV₁ was only 65% and 46% predicted, respectively. These differences between the four groups were both clinically relevant and statistically significant (p<0.001).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Cross sectional mean values of FEV₁ grouped by presence or absence of malnutrition and Pseudomonas aeruginosa (PA). The decline in % predicted FEV₁ at age 18 years is partly due to different reference values for adults.

Longitudinal analysis

Three patient subgroups were formed with respect to the patients' individual courses from 1995 to 1996. Malnourished adolescents who experienced a decrease of ≥5% predicted in weight for height during that year had a concomitant mean loss in FEV₁ of 16.5% predicted, whereas patients who gained relative weight had a parallel increase in FEV₁ of 2.1% predicted. These differences were statistically significant (p<0.001). In adolescents with normal weight in 1995, FEV₁ declined by 7.6% or 0.16% of predicted if weight decreased by at least 5% or remained stable, respectively, whereas patients who improved their weight by more than 5% had a concomitant increase in FEV₁ of 4.3% (p<0.001). Figure 3 shows the respective data for 6–11.9 year old children.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Mean values of individual differences in FEV₁ (% predicted) between 1995 and 1996 in 536 schoolchildren aged 6–11.9 years of age. (A) Normally nourished children. (B) Children who were malnourished in 1995 (weight for height (W/H) <90% predicted). If W/H decreased by 5% or more, a concomitant decline in FEV₁ of (A) 12.8% and (B) 7.5% predicted was observed.

In a further step a mixed model analysis was performed to identify differences between sexes and between risk groups (patients with or without malnutrition or P aeruginosa infection, respectively) with respect to the decline in FEV₁ within a year. Malnourished patients of all age groups had significantly worse lung function than their normally nourished counterparts (table 3). At the initial visit in 1995, FEV₁ was 11.7–23.9% predicted lower if weight for height was decreased . During the 2 year follow up period, adolescents with malnutrition experienced a decline in FEV₁ whereas lung function remained stable in adolescents with normal body weight (p<0.01, fig 4). The FEV₁ slopes of patients colonised with P aeruginosa were significantly worse in children and adolescents with a 1.14–2.79% predicted faster decline in lung function (p<0.001). The only association with sex was found in adolescents, with females showing a slower progression of lung disease than males—that is, a 1.58% predicted difference in FEV₁ within a year (table 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Mixed model analysis for the course of FEV₁ between 1995 and 1997 in the three age groups by risk factors (malnutrition and P aeruginosa (Pa) colonisation) and sex.

DISCUSSION

The negative impact of underweight on the long term outcome of patients with CF has long been recognised.³ The results of the present longitudinal analysis from a large cohort of patients confirm that nutrition and lung function are co-dependent variables in CF. Patients with normal weight had a significantly smaller decrease in lung function over a 2 year period than those with malnutrition, and this was shown in all age groups. A fall in weight for height of 5% predicted or more within 1 year was associated with a parallel decrease in FEV₁, whereas patients with improved nutrition showed constant or even improved FEV₁.

Cross sectional analysis revealed that the prevalence of malnutrition was 19% in children aged 2–6 years and increased considerably with age up to 38% in adults. Other disease related parameters, particularly Po₂ and MEF₂₅, were abnormal in even larger proportions of patients, and showed a similar deterioration with age. Of the 858 adolescents aged 12–18 years, more than one third had abnormal MEF₂₅ (59%), serum IgG (53%), Po₂ (50%), FEV₁ (42%), and VC (39%). The figures for FEV₁ are comparable to data from other patient registries. For example, the US registry found that FEV₁ was less than 70% predicted in 68% of adult patients.²¹ There are no data from other patient registries with respect to IgG, Po₂, or MEF₂₅ grouped by age.

In the present study, adolescent girls had better FEV₁ results and a smaller yearly decline in FEV₁ than boys. Younger children and adults showed no sex related differences by mixed model analysis. Data from the literature concerning the influence of sex on the course of CF are controversial. Zemel et al reported better weight (% predicted) in boys and higher Z scores for length and age in girls.¹³ In contrast, the growth of boys with CF was impaired on the basis of height, fat free mass, and fat mass when observed longitudinally.¹⁴ In the study by Corey et al, male patients had milder disease than females, as reflected by a smaller decline in FEV₁, decreased mortality, and higher survival.²² No significant differences between males and females have been found in other studies.^11,23–25

Chronic infection with P aeruginosa had a negative impact on pulmonary function, Po₂, and IgG. Significantly worse mean FEV₁ values were found in malnourished children, adolescents and adults, and the yearly decline in FEV₁ was larger in P aeruginosa positive patients. On the other hand, a large proportion of infected patients had normal weight and lung function. A recent report revealed a faster decline in FEV₁/FVC and an increased rate of worsening of chest radiographs after colonisation with P aeruginosa.²⁶ Previous studies found that infection with mucoid P aeruginosa was associated with an accelerated rate of decline in pulmonary function.^27,28 The CFQA project does not distinguish between mucoid or non-mucoid strains. From the data in the German registry, P aeruginosa infection must be regarded as a significant confounder for pulmonary function impairment in patients with CF.

When patients were stratified according to spirometric results, schoolchildren and adolescents with normal FEV₁ had a 4% and 10% predicted better mean weight for height, respectively, than children with impaired respiratory function. Zemel et al analysed patients with FEV₁ above or below 90% predicted and described significantly better data for weight and height in the group with normal lung function.¹³ The North American Epidemiologic Study of Cystic Fibrosis (ESCF) reported that mean weight for age was at the 45th percentile in patients with an FEV₁ of ≥100% predicted, whereas the mean weight for age percentile was only 20 if FEV₁ was significantly impaired (below 40% of normal).²⁵ Thus, a cross sectional perspective leaves no doubt that children with normal lung function have significantly better weight.

The detailed longitudinal follow up in the present study together with a mixed model analysis adds important information to previously published data. Corey et al²² discussed the advantages of a mixed model analysis—for example, that repeated observations are properly accounted for in significance testing. They described a decline of only 0.7% FEV₁ per year in patients with normal pancreatic function compared with a decrease of 3.1% in patients with pancreatic insufficiency. Since patients with pancreatic insufficiency generally have a lower body weight, this is indirect evidence of the influence of nutrition on the longitudinal development of lung function. Wasting has also been shown to be an independent predictor of mortality in a recent paper published in Thorax.²⁹ In the study by Zemel et al which included a four year longitudinal analysis, Z scores for weight for age and height for age were positively associated with FEV₁ % predicted.¹³ We have shown a 1.6% predicted smaller decline in FEV₁ in well nourished children and adolescents and a 12% difference in mean FEV₁ values between groups with or without malnutrition. When individual patients were categorised by change of weight and lung function within 1 year, those with a decrease in weight for height of ≥5% had a parallel decrease in FEV₁ and vice versa, whereas patients with constant or even improved weight experienced no or only an insignificant decrease in FEV₁. We cannot tell what the reasons for the changes in weight were since the CFQA project reports treatment by categories only. Clinical trials confirm, however, a parallel improvement in nutritional status and lung disease: intravenous antibiotics for respiratory exacerbations have a positive effect on both lung function and weight,^30–32 and inhaled antibiotics not only improve pulmonary function but also the nutritional status of patients.^33,34 On the other hand, long term gastrostomy feedings induce favourable outcomes of nutrition as well as of pulmonary function.³⁵

Possible reasons for the influence of malnutrition on chronic lung disease have been discussed in a comprehensive review.³⁶ In chronic obstructive pulmonary disease (COPD) emaciation has long been known to be accompanied by emphysema, and patients with COPD and severe airflow obstruction have lower body weight. Both oxygen consumption of respiratory muscles and resting energy expenditure were increased in patients with COPD. The flattened diaphragm present in emphysema diminishes the efficiency of respiratory muscles, and a myopathic process in malnutrition has also been suggested.³⁷ In animal studies biochemical abnormalities were associated with malnutrition—for example, increased collagen breakdown, impaired lung repair, and decreased surfactant activity. Relatively little information is available on possible pathophysiological processes in CF. In particular, a cause/effect relationship remains to be proven.⁵ In the malnourished mouse model of respiratory infections in CF, nutritional deficiency contributed to compromised innate lung defences, bacterial colonisation, and excessive inflammation in the CF respiratory tract.³⁸ A recent review summarised clinical studies on the relationship between lung function and nutritional status.³⁹ Correlations were found between FEV₁ and lean body mass, the body compartment which includes skeletal muscle, suggesting an influence of muscle wasting on pulmonary function. The detrimental effect of malnutrition on immune function in general⁴⁰ must also be considered as a relevant mechanism in CF. Treatment with intravenous antibiotics was associated with a concomitant decline in inflammatory parameters (C reactive protein, tumour necrosis factor-α), energy expenditure, and an increase in body weight.^32,41

In summary, the results of this study suggest that normal body weight is associated with better lung function in CF. A significantly smaller decline in FEV₁ can be expected in all age groups if patients are well nourished, and this is independent of a concomitant colonisation with P aeruginosa.